Daily numbers, graphs, analysis thread 13

[PatriciaHolm]

Taking the liberty of starting a new thread as we've just bust the old one, with much thanks to @BigChocfrenzy and I will copy her header..

Welcome to thread 13 of the daily updates

Resource links:

Slides & data UK govt pressers
UK dashboard sub-national data, local authorities
Beta Uk dashboard deaths, cases, hospitals, tests, partially sub-national
UK stats updated daily by PHE & DHSC
ONS UK statistics for CV related deaths, released weekly each Tuesday
PHE surveillance report infections & deaths released every Thursday with sep. infographic
NHS England stats including breakdown by Hospital Trust
FT Daily updates
HSJ Healthcare updates
Worldometer UK page
Plot FT graphs compare countries deaths, cases / million pop. / log / linear
Covidly.com filter graphs compare countries
Plot COVID Graphs Our World in Data

We welcome factual, data driven, and civil discussions from all contributors 📈📶👍

So, for example, the fact that many transport workers have died does rather suggest they caught it via work, or many of them did

Not necessarily. Their ethnicity, and therefore homelife and living conditions may be much more relevant. The recently commissioned reports will hopefully be able to pinpoint cause and effect etc. Within the NHS this is a big hot potato. Did staff catch it at work or bring it in to work....... lack of PPE either way is, of coures, the issue, but it is not a given that infection was acquired at place of work.

@cusano - I don't find it very helpful either. It lists cases per million, so I have to divide the number by 6 to start with (why per million when no area has 1M people?). I also know we reported something else a couple of times because obviously at the start we didn't know if it would turn out to be covid or not, but it didn't, and there's no way of saying "turns out DH had just eaten something dodgy" or "turns out DS has mild hayfever". It makes me question if I should add anything, so I don't know how well it can work.

Thanks @Firefliess that’s really helpful! So one day it said 112 per million and the next it said 1000 and then the day after said 650 etc I was so confused why or how it would go up and down so fast

@MRex yes! I agree here too! My area has 100,000 people so sometimes it looks alarming! And yes agree it’s actually not that helpful, I was logging some symptoms on April before I had a test (which was negative) so I guess before I had that result I’d have been classed as symptomatic

@Baaaahhhhh

So, for example, the fact that many transport workers have died does rather suggest they caught it via work, or many of them did

Not necessarily. Their ethnicity, and therefore homelife and living conditions may be much more relevant. The recently commissioned reports will hopefully be able to pinpoint cause and effect etc. Within the NHS this is a big hot potato. Did staff catch it at work or bring it in to work....... lack of PPE either way is, of coures, the issue, but it is not a given that infection was acquired at place of work.

SHoots analysed this and iirc it looks like occupation is far more important than ethnicity Look at death rates for these jobs in areas with a very low % BAME and hence where transport workers are nearly all white Also look at countries or regions that are more homogeneus ethnically

Jobs where people come into contact with a lot of others all day;
jobs where people are in close quarters with others all day
are higher risk

e.g. meat processing plants, staffed by whites (EE) with the added problem of v cold temps
fruit & veg pickers (EE)

[quote MRex]@cusano - I don't find it very helpful either. It lists cases per million, so I have to divide the number by 6 to start with (why per million when no area has 1M people?). I also know we reported something else a couple of times because obviously at the start we didn't know if it would turn out to be covid or not, but it didn't, and there's no way of saying "turns out DH had just eaten something dodgy" or "turns out DS has mild hayfever". It makes me question if I should add anything, so I don't know how well it can work.[/quote]
They're not assuming all relevant symptoms must be covid; that's not how it works. What they want to track are symptoms, then use random testing to see which symptoms are most likely to correspond to positive tests (and what fraction of people having particular symptoms turn out to be positive - also what sets of symptoms reported at the same time as each other or in a particular order tend to be correlated with positive tests).

To learn that, they need people to report all new symptoms, because to find out that X% of people reporting a sore throat combined with a headache for two days are likely to test positive, they also need to know how many people had that symptom combination without testing positive, too.

Then, as time goes by, they can start to make predictions about infection rates based on the fact that in a particular area a particular symptom set in a particular order might have gone up by 50%. So reporting symptoms isn't just about reporting probable covid, it's also about providing data about symptoms regardless of cause, to help the models improve.

Of course, people-facing jobs were a lot higher relative risk before masks came into use and SD measures were improved

Re asymptotic tests, I think this gives some small insight into the testing and tracking ( not the right word ) how the virus is progressing. I presume blanket testing in areas with clusters will show up more of those with no symptoms Hence the alert in France where this number is falling. So again presume people are asking for a test because of symptons rather than being found and this is rising.

@eeeyoresmiles - I understand all that in principle, and we are continuing to log. If you look at the rapid changes on their local area maps though, it's fairly clear that the model goes a bit mad. Which is ok, I prefer to follow confirmed cases and deaths anyway, but it would be nice if the model took a more measured approach in swinging both up and down.

I hope that there are long-term changes from all these studies, because clearly flu and norovirus spread every year; covid being more obvious has just exposed what are likely to be the same transmission routes through hospitals, care homes and vulnerable homes.

763 cases and 83 deaths today

7 day rolling average:

8/7 545
15/7 586
22/7 638
29/7 725

Creep, creep

I downloaded the figures and Bradford caught my eye, every time I spotted a highish figure ( in the 20s or 30s) it was Bradford. Looking at the beta graph, that's not really reflected yet. I haven't charted it out, so it could be my local bias as I live in Yorkshire!

@HoldingTight

7 day rolling average:

8/7 545
15/7 586
22/7 638
29/7 725

Creep, creep

National infection rates are no longer that important

We are testing lots more people and the increase numbers is almost entirely in the pillar 2 tests. We are picking up more younger people testing positive and more asymptomatic and mild cases. In addition There are huge regional variations.

Hospital admissions are falling and deaths are, at worst, plateauing.

I haven't got all the figures but I checked the testing figures one day last week and it was under 100,000. Today is also under 100,000. I'm not convinced we are testing more.

Does anyone have an easy to read link to the testing figures please?

Hospital admissions are definitely still falling quite a bit. Which means if we're not testing more, maybe we're testing in a more targeted fashion - ie testing everyone in a neighbourhood or who works somewhere where there's been an outbreak - ie getting better at picking up more cases.

But also possible that cases are rising in the general community, and that the reason hospital admissions is falling is that the outbreak in care homes is now under control. This is not a great scenario because if cases are rising, they'll go on rising unless we do something to stop that, and we will then eventually start to see hospital admissions rising back up again.

Purplecalm here is a link to testing data here. Testing numbers are a bit up and down each day but the trend on hospitalisations continue downwards so you’d hope that the uptick in cases is due to more focused testing in potential/actual hotspots so they will be picking up more pillar 2 cases that are milder or asymptomatic (and likely younger people).

X post Firefliess- good point about care homes vs community cases. Is there any resource that shows care home data?

@sleepwhenidie

Purplecalm here is a link to testing data here. Testing numbers are a bit up and down each day but the trend on hospitalisations continue downwards so you’d hope that the uptick in cases is due to more focused testing in potential/actual hotspots so they will be picking up more pillar 2 cases that are milder or asymptomatic (and likely younger people).

Can you do a link to hospital admissions too please? I can't find them even though I'm sure I've looked at them in the past.

Found this on care homes but it details outbreaks (which are decreasing) rather than cases...

Hospital data

Thank you

This is the tests (p1&p2) by date announced alongside % rate (for cases announced that day - not a perfect match but there is lag on both so this is the simplest and quickest proxy).

Also 7 day average on total P1&p2 tests and 7 day average for % rate.

ATM - too soon to say really! tests up until last few days, % rate a little up but still need more data to really see anything at a UK level.

Weekend US figures looking like they were a blip - over 1200 deaths yesterday and a new record for Florida today of 216 deaths, up from previous high of 186

Would a real increase in cases show up in hospital data if those new cases were more in younger people rather than older people?

It would show up eventually if more older people were eventually infected by those younger people, but perhaps not as quickly? What is the average day of hospital admission, is it still about day 10?

So the two daily figures published by the government - positive tests and deaths - are actually both less useful than the unannounced third stat, hospitalisations....